Pembrolizumab and chemotherapy in high-risk, early-stage, ER+/HER2- breast cancer: a randomized phase 3 trial

Abstract

Addition of pembrolizumab to neoadjuvant chemotherapy followed by adjuvant pembrolizumab improved outcomes in patients with high-risk, early-stage, triple-negative breast cancer. However, whether the addition of neoadjuvant pembrolizumab to chemotherapy would improve outcomes in high-risk, early-stage, estrogen receptor-positive/human epidermal growth factor receptor 2-negative (ER⁺/HER2^-) breast cancer remains unclear. We conducted a double-blind, placebo-controlled phase 3 study (KEYNOTE-756) in which patients with previously untreated ER⁺/HER2^- grade 3 high-risk invasive breast cancer (T1c-2 (≥2 cm), cN1-2 or T3-4, cN0-2) were randomly assigned (1:1) to neoadjuvant pembrolizumab 200 mg or placebo Q3W given with paclitaxel QW for 12 weeks, followed by four cycles of doxorubicin or epirubicin plus cyclophosphamide Q2W or Q3W. After surgery (with/without adjuvant radiation therapy), patients received adjuvant pembrolizumab or placebo for nine cycles plus adjuvant endocrine therapy. Dual primary endpoints were pathological complete response and event-free survival in the intention-to-treat population. In total, 635 patients were assigned to the pembrolizumab-chemotherapy arm and 643 to the placebo-chemotherapy arm. At the study's prespecified first interim analysis, the pathological complete response rate was 24.3% (95% confidence interval (CI), 21.0-27.8%) in the pembrolizumab-chemotherapy arm and 15.6% (95% CI, 12.8-18.6%) in the placebo-chemotherapy arm (estimated treatment difference, 8.5 percentage points; 95% CI, 4.2-12.8; P = 0.00005). Event-free survival was not mature in this analysis. During the neoadjuvant phase, treatment-related adverse events of grade ≥3 were reported in 52.5% and 46.4% of patients in the pembrolizumab-chemotherapy and placebo-chemotherapy arms, respectively. In summary, the addition of pembrolizumab to neoadjuvant chemotherapy significantly improved the pathological complete response rate in patients with high-risk, early-stage ER⁺/HER2^- breast cancer. Safety was consistent with the known profiles of each study treatment. Follow-up continues for event-free survival. ClinicalTrials.gov identifier: NCT03725059 .

Fig. 1. Disposition of patients in the study.

Progressive disease included radiographic progressive disease. Patients did not have to complete all neoadjuvant therapy to undergo surgery.

Fig. 2. Subgroup analysis of the difference in percentages of patients with a pCR at the first interim analysis.

Data from key subgroups are shown. For the overall population, the estimated mean treatment difference with 95% CI was calculated using the stratified Miettinen–Nurminen method. The analysis for PD-L1 CPS subgroups was stratified. All other analyses were unstratified. The footnote ‘a’ indicates the subgroup analyses of pCR based on a PD-L1 CPS cutoff of 20 were not prespecified. PD-L1 CPS was defined as the number of PD-L1-positive tumor cells, lymphocytes and macrophages divided by the total number of tumor cells multiplied by 100. ECOG performance status ranges from 0 to 5, with higher scores indicating greater disability.

Fig. 3. RCB at the first interim analysis.

RCB was an exploratory endpoint and was assessed by a local pathologist at the time of surgery. RCB-0, RCB-1, RCB-2 and RCB-3 denote increasingly larger residual diseases. RCB data were missing for 24 (3.8%) and 14 patients (2.2%) in the pembrolizumab‒chemotherapy and placebo‒chemotherapy arms, respectively.

Extended Data Fig. 1. Post hoc exploratory analysis of the difference in percentages of patients with a pathological complete response by estrogen receptor positivity and PD-L1 combined positive score at the first interim analysis.

Programmed cell death ligand 1 (PD-L1) combined positive score (CPS) was defined as the number of PD-L1-positive tumor cells, lymphocytes and macrophages divided by the total number of tumor cells multiplied by 100. No pathological complete response occurred in patients with tumor PD-L1 CPS < 1 with estrogen receptor (ER) positivity <10% (pembrolizumab arm, n = 1; placebo arm, n = 4). For the overall population, the estimated treatment difference was calculated using the Miettinen–Nurminen method stratified by region (China vs Eastern Europe vs all other countries) and PD-L1 status (CPS ≥ 1 vs <1). Estimated treatment differences for other endpoints were based on the Miettinen–Nurminen method (unstratified).

Extended Data Fig. 2. Multiplicity graph for type I error control of study hypotheses.

The prespecified analysis plan allows alpha passing from successful endpoint(s) to other(s). H, hypothesis.